Polyethylene wear rate and osteolysis: critical threshold versus continuous dose-response relationship

Transcription

1 ELSEVER Journal of Orthopaedic Research 23 (2005) Journal of Orthopaedic Research Polyethylene wear rate and osteolysis: critical threshold versus continuous dose-response relationship J. Mark Wilkinson Andrew J. Hamer a, an Stockley a, Richard Eastell a Lower Limb Arthroplusty Unit, Depurtnwnt of Ortltopuedics, Northern Givwul Hospitul. fferries Road, Shefield, S5 7A U. United Kingdom Acurlemic Unit of Bone Metubolisnl, Dillision of Clinicul Sciences (North), University of Shefield, Shefield, United Kingdom Received 6 April accepted 15 November 2004 Abstract We studied the relationship between polyethylene wear and osteolysis in 230 subjects after cemented Charnley total hip arthroplasty in order to examine the validity of the wear rate threshold concept. Polyethylene wear measured using image analysis (EBRA) software was compared in 11 5 subjects with osteolysis versus 115 control subjects that were individually matched for age, sex, and follow up period. Subjects with osteolysis had almost twice the mean annual wear rate versus the controls. The incidence of osteolysis increased in a linear manner with each quintile increase in wear rate throughout the range mm/year. The odds-ratio for osteolysis approximately doubled with each quintile increase in wear rate above the middle quintile (wear rate m d year), and decreased at a similar rate with each quintile decrease in wear rate below the middle quintile. Our data suggests that the association of osteolysis with polyethylene wear rate represents a continuous dose-response relationship and does not support the concept of a discrete critical wear rate threshold above which the risk of osteolysis is disproportionately increased Orthopaedic Research Society. Published by Elsevier Ltd. All rights reserved. Keyivords: Total hip arthroplasty; Osteolysis; Polyethylene wear; EBRA ntroduction Data from several clinical studies have suggested that high bearing surface polyethylene wear rate and total wear are risk factors in the development of osteolysis and aseptic loosening after total hip arthroplasty (THA) [6,2 1,23,24]. These findings are consistent among studies of various implant designs and using different wear measurement techniques [7,20]. The results of these studies have led some investigators to suggest a critical wear rate threshold above which the risk of osteolysis is increased [6,7,21]. The wear rate threshold concept has several implications for implant design, patient monitoring, and under- * Corresponding author. Tel.: ; fax: E-muil uddress: wilkomark@aol.com (J.M. Wilkinson). standing of the biological response to particulate debris. t implies that polyethylene wear is the dominant risk factor for osteolysis. t suggests that above a certain wear rate there is a disproportionately increased osteolytic response. Thus if an implant s wear rate characteristics perform below this threshold it s long-term survival is likely to be good. The practical application of such a threshold as a screening tool for at-risk patients also implies that inter-individual difference in biological responsiveness to particulate debris does not impact significantly on the threshold value. The validity of the wear rate threshold concept remains controversial. Harris recently identified several limitations to the supporting literature [lo]. These included short follow-up, unclear definitions of osteolysis, the observation that osteolysis does occur in individuals with low rates of polyethylene wear, limitations of plain radiographs for defining osteolysis, and uncontrolled effects of other variables /$ - see front matter Orthopaedic Research Society. Published by Elsevier Ltd. All rights reserved. doi: /j.orthres

2 J. M. Wilkinson et trl. 1 Journul q/ Ortliopuedic Resrurcli 23 (2005) An additional problem of examining the relationship between wear rate and osteolysis in cohort studies is that the endpoint of interest is relatively uncommon. Thus the power to study the relationship using this type of study design may be limited unless the sample size is very large or the incidence of osteolysis atypically high. Differences in underlying disease, age, sex, and follow up period may also confound analysis, as groups of subjects with osteolysis will typically also tend to exhibit these risk factors. Finally, studies that have quantified the relationship between wear rate and osteolysis have usually done so by examination of the data after division of the subjects into wear rate groups based on fixed wear rate increments (Fig. 1) [21,24]. However, wear rate data tend to be distributed in a skewed manner, with a long right-sided tail [22]. This results in disproportionate clustering of subjects into the first and second wear rate intervals, while the numbers of subjects in successive intervals declines. This method demonstrates that increasing wear is associated with a greater incidence of osteolysis. However, it lacks sensitivity in addressing the question that arises from the critical wear threshold 50 Equal interval by wear rate _ \ Annual linear wear (nun) Equal interval by subject numbers lo-.. - l l Annual linear wear (mm) Fig. 1. Distribution of wear rates within a given population divided into groups based on fixed wear rate interval (top panel, number of subjects in each group uneven), versus division by wear quintile (bottom panel, number of subjects in each group equal). concept, that is: how low does the wear rate need to be to eliminate osteolysis? The aim of this study was to quantify the relationship between linear polyethylene wear rate and osteolysis in an individually case-matched association study of 230 subjects after cemented Charnley THA. By choosing this study design we aimed to control for disease, implant, sex, age at surgery, and follow up period on a case-by-case basis, and also include a large number of osteolysis subjects. We specifically aimed to determine whether there is evidence to support the concept of a critical threshold value by dividing subjects into wear rate quintiles (thus ensuring an equal number of subjects in successive wear rate groups) and examining the change in number of subjects with osteolysis between quintiles (Fig. 1). Patients and methods Subjects were recruited between February 2000 and February and included Caucasian men and women who had previously undergone THA for idiopathic osteoarthritis of the hip between the years 1973 and All subjects received a cemented monobloc Charnley femoral component with a 22.2 nun diameter femoral head, and a cemented Charnley polyethylene acetabular component. The study subjects comprised a matched subgroup from a larger study of 481 subjects examining risk-factors for osteolysis after cemented THA [28]. Subjects were excluded if they had any history of inflammatory arthropathy, or known secondary causes of hip arthritis such as trauma, avascular necrosis, and developmental or childhood hip disease. Subjects were also excluded if they had taken courses of immunosupressant agents or bisphosphonates for a continuous period of greater than six months since THA, or if there was clinical suspicion of implant infection. All subjects provided written informed consent prior to participation. The study was approved by North Sheffield Local Research Ethics Committee and conducted in accordance with the ethical principles stated in the Declaration of Helsinki. The osteolysis group consisted of subjects with aseptic loosening diagnosed on plain radiographs of the hip taken prior to revision surgery and had loosening confirmed surgically, and subjects who had osteolytic lesions or established aseptic loosening on current plain antero-posterior and lateral radiographs. Loosening of the prosthetic stem and cup were defined according to the criteria of Harris and McGann [ll] and Harris and Penenberg [12], respectively, and included those implants classified as definitely or probably loose. Femoral and pelvic bone defects were classified according to the American Academy of Orthopaedic Surgeons (AAOS) grading system [2,3]. The control group comprised subjects with no evidence of aseptic loosening or osteolytic lesions on plain antero-posterior and lateral radiographs of the hip taken on the day of clinical review. The criteria for individual matching of osteolysis subjects and controls were sex, age at primary THA (within 5 years) and follow-up period (within 2 years). Radiographs were digitized using a laser digitizer (Lumiscan 200, Lumisys nc., Sunnyvale, CA). Acetabular cup polyethylene linear wear in the antero-posterior plane was measured using EBRA software (EBRA-Cup, release 2000, University of nnsbruck, Austria) [27]. Wear was measured as the difference in centering of the femoral head and the center of the cup using a supine AP radiograph of the pelvis. For the control subjects the measured radiograph was that taken on the day of clinical assessment. n the case of the osteolysis subjects the last radiograph taken prior to revision surgery was used. Polyethylene wear was expressed as total linear wear and as mean annual linear wear rate. Mean annual wear rate was calculated as total linear wear divided by time since primary surgery. Results are presented as mean f standard deviation for normally distributed data and median +_ interquartile ranges for nonnormally distributed data. Between-group comparisons for

3 522 J.M. Wilkinson et al. Journal of Orthvpuedic Research 23 (2005) individually-matched continuous variables were made using the paired 1-test or the Wilcoxon test, as appropriate. Comparisons of nonmatched continuous data were made using the Mann-Whitney test. Categorical variables and differences in the number of subjects with osteolysis between wear rate quintiles were analysed using the X'-test. Differences in osteolysis rate between quintiles were also expressed as odds-ratios. All analyses were 2-tailed using a critical P value of 0.05, and made using SPSS statistical software (version 12, SPSS UK, Chertsey, U.K.). Results Subjects in the osteolysis group (n = 115) were of similar age at primary surgery, sex, height, weight, body mass index, and follow-up period as those in the control group (Table 1; P > 0.05 all comparisons). Eighty-eight subjects (76.5%) had aseptic loosening confirmed at revision surgery. n the remainder revision had been declined or was not clinically indicated. The distribution of the wear data was log-normal. Median (interquartile range) total linear wear was 1.28 mm ( ) and 0.74 ( ) in the osteolysis and control groups, respectively (Wilcoxon, z = -5.5, P < 0.001). Median annual linear wear was 0.12 mm ( ) and 0.07 ( ) in the osteolysis and control groups, respectively (Fig. 2, Wilcoxon, , P < 0.001). n the osteolysis group 22 subjects -- had osteolysis around the pelvic component only, in 30 the femoral component only was involved, and in 63 there was osteolysis around both implant components. Subjects with osteolysis around the pelvic component, or around both the pelvic and femoral components had a higher annual wear rate than those with only femoral osteolysis (Fig. 2, Mann-Whitney, P < 0.001, and P = 0.02, respectively). The distribution of AAOS grades in the 85 subjects with pelvic osteolysis was 32, 8, 27, and 18 for grades 0 through 3, respectively (where grade 0 = linear osteolysis only). The distribution of the AAOS grades in the 93 subjects with femoral osteolysis was 31, 4, 48, and 10 for grades 0 through 3, respectively. Subjects were divided into wear rate quintiles (ql-q5) based on annual linear wear rate (n = 46 subjects per Table 1 Characteristics of study subiects Pa tien t characteristic THA osteolysis (n = 1 S) THA controls (n = 1 S) "Age at primary THA (years) 61.0 f f 7.5 bsex (mae:femae) 49:66 49:66 "Height (m) 1.66 f f 0.09 "Weight (kg) 76.6 f ? 15.5 "Body mass index (kg/m') 27.6 f f 4.9 "Follow up period (years) 11.0 f f2.9 Plus-minus figures are mean f SD. P > 0.05 all comparisons. a Paired?-test. x2 test with Yates' correction a, c b b a, ' 0.3-1z m C m 2.E P) c P-=O.OO t ' no osteolysis osteolysis 0.61 P=0.02 P - P<O.OO c------, 0.34 V.V pelvis femur both Fig. 2. Box and whisker plots showing annual linear wear in osteolysis and control subjects (top panel), and annual linear wear in subjects with osteolysis of the pelvis, femur, or of both pelvis (bottom panel, broken line indicates median annual wear rate in control subjects). Bar indicates median, box indicates interquartile range, whiskers show range. Statistical analyses are by Wilcoxon test and Mann-Whitney test, respectively. quintile, Fig. 3). The lower and upper boundaries for the quintiles (units = mdyear) were (ql), (q2), (q3), (q4), and (q5). The number of subjects with osteolysis was significantly different between the quintile groups (x' = 35.6, P < 0.001). The number of subjects with osteolysis in each successive quintile group increased in a linear manner. The odds-ratio for osteolysis for quintiles 1, 2, 4, and 5 were expressed in comparison to the proportion of subjects with osteolysis in the middle quintile (q3). There was an almost linear increase in the odds of osteolysis with increasing wear rate quintile, ranging from 0.3 for quintile 1 up to 4.9 for quintile 5 (Fig. 3). An alternate approach to quantifying the relationship between wear rate and osteolysis is to use logistic regression analysis, and to express the results as odds of osteolysis per unit change in wear [21]. Using this method (and the median wear rate in the controls as the incremental unit), the odds ratio for osteolysis for each 0.07mdyear change in wear rate was 2.4 (logistic regression, 95% confidence interval , P < 0.001). Using the standard deviation (SD) as the incremental unit (data log transformed) the odds-ratio for each SD change in wear rate was 2.2 (95% confidence interval , P < 0.001).

4 J. M. Wilkinson et al. Journal ($ Orthoprimdie Research 23 (2005) u).- % 40 b S, 30 0 & n 20 5 = 10 n. EXQ osteolysis 0 no osteolysis x2 test, P<O.OO wear quintile wear quintile Fig. 3. Histogram showing number of subjects with osteolysis by quintile of annual linear wear (top panel, n = 46 subjects per quintile). Bottom panel shows odds-ratio +95% confidence interval for osteolysis by quintile group versus quintile 3. Discussion n this case-matched association study we aimed to characterise the relationship between linear wear rate and osteolysis. We found that the number of subjects with osteolysis increased with wear rate quintile throughout the range mdyear. The oddsratio for osteolysis, which is similar to the risk of osteolysis, approximately doubled with each quintile increase in wear rate above quintile 3 (wear rate mm/ year). The odds-ratio also decreased at a similar rate for each quintile decrease in wear rate below quintile 3. These findings suggest that the relationship between wear rate and osteolysis represents a continuous doseresponse relationship for the Charnley prosthesis. The wear rate threshold concept implies that wear debris is the dominant factor that predisposes to osteolysis and aseptic loosening, and that there is a saturable mechanism for neutralisation or removal of wear debris. Once this level is exceeded the risk of osteolysis is disproportionately increased. t also suggests that inter-individual biological response to debris is limited. However, data from several studies suggest that other factors such as early implant stability, bone geometry, and underlying disease also modulate aseptic loosening [8,14,16]. n-vitro studies also suggest that the biological response to debris differs between individuals, and between alleles for a particular gene [9,18]. Genetic differences between individuals are also associated with differences in risk of osteolysis in-vivo [28]. These observations are inconsistent with the concept of a discrete wear rate threshold, but remain consistent with the concept that increasing wear rate increases the risk of osteolysis. Our quintile-based analysis approach to the doseresponse relationship contrasts with previous studies that have used fixed wear-rate increments. We feel that this method allows more consistent analysis throughout the wear rate range given equal numbers of subjects in each group. This approach has been used extensively in previous epidemiological studies, for example in quantifying the relationship between bone mineral density and risk of hip fracture [l], blood pressure and outcome after stroke [4], and between serum cholesterol levels and myocardial infarction [ 191. However, subject grouping by wear rate quintile does result in differences in the ranges of wear rate contained within each quintile. Thus, although the dose-response relationship is continuous, it is not necessarily linear. Our data reflect the clinical observation that osteolysis occurs, albeit uncommonly, in the presence of a low polyethylene wear rate. Dumbleton et al. have suggested that a wear rate threshold of 0.05 mdyear linear wear rate would, for practical purposes, eliminate osteolysis [7]. However, 10 of our subjects with osteolysis (9%) had a wear rate that was lower than this threshold value. The mean annual wear rate of both the osteolysis subjects and controls in our study was similar to that found by Sochart in a cohort study of 235 patients with Charnley implants at a mean follow up of 19.5 years ~41. Our finding that wear rates were different between subjects with pelvic versus femoral osteolysis was unexpected, but consistent with previous reports of Charnley implants. Kesteris et al., in a study of 74 revised implants found a lower wear rate in subjects with femoral versus pelvic osteolysis (0.1 mdyear versus 0.3 mm/ year, respectively; P = 0.001) [15]. Sochart also reported a weaker relationship between wear rate and femoral component loosening than that between wear rate and pelvic component revision [24]. n the design of this study we aimed to address some of the limitations of some previous studies by using a large sample size and including and equal number of osteolysis subjects and controls. All subjects had the same indication for surgery and all received the same implant. We also aimed to control for physical characteristics and follow up time by individually matching disease patients to controls. We have focussed on the relationship between wear rate and osteolysis as the threshold

5 524 J.M. Wilkinson et al. Journal of Orthopuedic Reseurch 23 (2005) concept is commonly expressed in these units. There is also an established relationship between total wear and osteolysis [24,29]. However, to examine this relationship would require a wide range of follow up periods between subjects that might introduce unrecognised biases between subjects. For example, changes in implant manufacturing and sterilising techniques, alterations in operating practice and rehabilitation programs, as well the types and levels of activity to which subjects expose their implants may confound interpretation of the effect of total wear in subjects where a very broad range of follow up periods is involved. Our observations were made using the Charnley implant, and cannot necessarily be generalised to other prosthesis designs. We used two-dimensional wear measurements. These may underestimate wear when compared to three-dimensional methods [ 13,261, although they have the advantage of being simpler and more reproducible [17]. Previous studies have also shown linear wear measurements have similar predictive value for osteolysis as volumetric wear measurement [211. We also measured wear using a single radiograph, and calculated the mean wear rate based on the time since implantation. This assumes that the femoral head and cup were concentric at time zero and the wear rate is constant. During the first 2 years measured wear rates are higher due to the processes of bedding-in and creep [5]. After this period the wear rate for individual implants is generally constant [6,25]. Our calculated wear rates are thus likely to be slightly higher than the true wear rates because of the initial bedding-in and creep components. However, this effect is likely to have been similar in both osteolysis and control subjects as follow up period was matched on a case-by-case basis. n summary our data suggests that the association between linear wear rate and osteolysis represents a continuous dose-response relationship and does not support the wear threshold concept. This data suggests that aims to improve the wear characteristics of implant materials should aim for the lowest wear rate possible. However, this may not completely eliminate osteolysis, particularly of the femur, as other factors may also contribute to its development. Acknowledgement This study was funded by a grant from the Arthritis Research Campaign. References [] Cummings SR. Black DM, Nevitt MC, Browner W, Cauley J, Ensrud K. et al. Bone density at various sites for prediction of hip fractures. The study of Osteoporotic Fractures Research Group. Lancet 1993;341:72-5. [2] DAntonio J, McCarthy JC. Bargar WL, Borden LS. Cappelo WN, Collis DK, et al. Classification of femoral abnormalities in total hip arthroplasty. Clin Orthop 1993;296: [3] DAntonio JA, Capello WN, Borden LS, Bargar WL, Bierbaum BF, Boettcher WG, et al. Classification and management of acetabular abnormalities in total hip arthroplasty. Clin Orthop 1989;243: [4] Dawson SL, Manktelow BN, Robinson TG. Panerai RB. Potter JF. Which parameters of beat-to-beat blood pressure and variability best predict early outcome after acute ischemic stroke? Stroke 2000;31: [5] Devane PA, Horne JG. Assessment of polyethylene wear in total hip replacement. Clin Orthop 1999;369: Dowd JE, Sychterz CJ, Young AM, Engh CA. Characterization of long-term femoral-head-penetration rates. Association with and prediction of osteolysis. J Bone Joint Surg 2000;82-A: [7] Dumbleton JH, Manley MT, Edidin AA. A literature review of the association between wear rate and osteolysis in total hip arthroplasty. J Arthroplasty 2002; 17: [8] Furnes 0. Lie SA. Espehaug B. Vollset SE, Engesaeter LB, Havelin L. Hip disease and the prognosis of total hip replacements. A review of 53,698 primary total hip replacements reported to the Norwegian Arthroplasty Register J Bone Joint Surg 2001;83-B: [9] Gordon A. Kiss-Toth E, Wilson AG, Eastell R. Wilkinson JM. A single nucleotide polymorphism at -238 in the TNF gene influences promoter response to polyethylene particles in-vitro. Calcif Tissue nt 2003;72:259. [lo] Harris WH. The lysis threshold : an erroneous and perhaps misleading concept? J Arthroplasty 2003; 18: [ ] Harris WH, McGann WA. Loosening of the femoral component after use of the medullary-plug cementing technique. Follow-up note with a minimum five-year follow-up. J Bone Joint Surg 1986;68-A: [21 Harris WH. Penenberg BL. Further follow-up on socket fixation using a metal-backed acetabular component for total hip replacement. a minimum 10-year follow-up study. J Bone Joint Surg 1987;69-A [31 Hui AJ, McCalden RW, Martell JM. MacDonald SJ, Bourne RB. Rorabeck CH. Validation of two and three-dimensional radiographic techniques for measuring polyethylene wear after total hip arthroplasty. J Bone Joint Surg 2003;85-A: [14] Karrholm J, Borssen B, Lowenhielm G, Snorrason F. Does early micromotion of femoral stem prostheses matter? 47 year stereoradiographic follow-up of 84 cemented prostheses. J Bone Joint Surg 1994;76-B: [51 Kesteris U, Hardinge K. llchmann T, Wingstrand H. Polyethylene wear in prosthetic hips with loose components. J Arthroplasty 2003; 8 : [61 Kobayashi S. Takaoka K, Saito N. Hisa K. Factors affecting aseptic failure of fixation after primary Charnley total hip arthroplasty. Multivariate survival analysis. J Bone Joint Surg 1997;79-A [71 Martell JM, Berkson E. Berger R, Jacobs J. Comparison of two and three-dimensional computerized polyethylene wear analysis after total hip arthroplasty. J Bone Joint Surg 2003;85- A: [81 Matthews JB, Green TR, Stone MH, Wroblewski BM, Fisher J. ngham E. Comparison of the response of primary human peripheral blood mononuclear phagocytes from different donors to challenge with model polyethylene particles of known size and dose. Biomaterials 2000;21: [91 Neaton JD, Wentworth D. Serum cholesterol, blood pressure. cigarette smoking, and death from coronary heart disease. Overall

6 J. M. Wilkinson E a/. Journcil of Ortliopaedic Reseurch 23 (2005) findings and differences by age for 316,099 white men. Multiple risk factor intervention trial research group. Arch nt Med 1992; 152:5&64. [20] Oparaugo PC, Clarke C, Malchau H, Herberts P. Correlation of wear debris-induced osteolysis and revision with volumetric wearrates of polyethylene: a survey of 8 reports in the literature. Acta Orthop Scand 2001;72:22-8. [21] Orishimo KF, Claus AM, Sychterz CJ, Engh CA. Relationship between polyethylene wear and osteolysis in hips with a secondgeneration porous-coated cementless cup after seven years of follow-up. J Bone Joint Surg 2003;85-A: [22] Pedersen DR, Brown TD, Hillis SL, Callaghan JJ. Prediction of long-term polyethylene wear in total hip arthroplasty, based on early wear measurements made using digital image analysis. J Orthop Res 1998;16: [23] Perez RE, Rodriguez JA, Deshmukh RG, Ranawat CS. Polyethylene wear and periprosthetic osteolysis in metal-backed acetabular components with cylindrical liners. J Arthroplasty 1998; 13:l-7. [24] Sochart DH. Relationship of acetabular wear to osteolysis and loosening in total hip arthroplasty. Clin Orthop 1999;363: [25] Sychterz CJ, Engh Jr CA, Yang A, Engh CA. Analysis of temporal wear patterns of porous-coated acetabular components: distinguishing between true wear and so-called bedding-in. J Bone Joint Surg 1999;81-A: [26] Sychterz CJ, Yang AM, McAuley JP, Engh CA. Two-dimensional versus three-dimensional radiographic measurements of polyethylene wear. Clin Orthop 1999;365: [27] Wilkinson JM. Hamer AJ, Elson RA, Stockley, Eastell R. Precision of EBRA-Digital software for monitoring implant migration after total hip arthroplasty. J Arthroplasty 2002; 17: [28] Wilkinson JM, Wilson AG, Stockley 1, Scott R. Macdonald DA. Hamer AJ, et al. Variation in the TNF gene promoter and risk of osteolysis after total hip arthroplasty. J Bone Miner Res 2003; 18: [29] Wroblewski BM, Siney PD. Charnley low-friction arthroplasty in the young patient. Clin Orthop 1992;285:45-7.